Re: [PATCH net] tun: Remove nested call to bpf_net_ctx_set() in do_xdp_generic()

From: Jeongjun Park
Date: Thu Jul 25 2024 - 08:15:28 EST


Paolo Abeni wrote:
>
> On 7/25/24 04:43, Willem de Bruijn wrote:
> > Jeongjun Park wrote:
> >> In the previous commit, bpf_net_context handling was added to
> >> tun_sendmsg() and do_xdp_generic(), but if you write code like this,
> >> bpf_net_context overlaps in the call trace below, causing various
> >> memory corruptions.
> >
> > I'm no expert on this code, but commit 401cb7dae813 that introduced
> > bpf_net_ctx_set explicitly states that nested calls are allowed.
> >
> > And the function does imply that:
> >
> > static inline struct bpf_net_context *bpf_net_ctx_set(struct bpf_net_context *bpf_net_ctx)
> > {
> > struct task_struct *tsk = current;
> >
> > if (tsk->bpf_net_context != NULL)
> > return NULL;
> > bpf_net_ctx->ri.kern_flags = 0;
> >
> > tsk->bpf_net_context = bpf_net_ctx;
> > return bpf_net_ctx;
> > }
>
> I agree with Willem, the ctx nesting looks legit generally speaking.
> @Jeongjun: you need to track down more accurately the issue root cause
> and include such info into the commit message.
>
> Skimming over the code I *think* do_xdp_generic() is not cleaning the
> nested context in all the paths before return and that could cause the
> reported issue.

Thanks to your comment, I re-read the code and found the root cause.
I will send a patch for that bug.

Regards,
Jeongjun Park